Which object do I receive?
Field names in each table are relative to that object. For example,
account.domain on a Lead is the domain field in AccountSummary. An embedded AccountSummary contains basic company identity, not every field on the Account response.
JSON types and missing values
Successful list responses always include a
data array and a cursor object. A final page has cursor.next: "" and cursor.has_more: false. All object fields shown in successful response models are emitted; optional objects and timestamps use null instead of disappearing.
On Leads and Accounts, an uncalculated evaluation.ai_score is null; on a finding, ai_score is null. If the record has no evaluation, evaluation itself is null. A calculated score of zero is an object containing value: 0; it is not missing data. Its fit matches the portal: weak from 0 through 29, low from 30 through 49, medium from 50 through 79, and high from 80 through 100. Only records with an available score can match an explicit AI score range filter.
The downloadable OpenAPI specification uses the Swagger 2.0 x-nullable extension to express nullable fields. The Nullable column below describes the actual JSON responses. Schema examples are illustrative; they do not guarantee that a field will be populated for every record.
Records, findings, and counts
Lead and Account lists return one row per Lead or Account.evaluation contains the selected saved evaluation and up to five findings linked to that exact evaluation. The assessment is returned only inside evaluation. source_sheet is the selected source context, rather than every Sheet membership. Use evaluation.findings_cursor with evaluation_id on signal history to continue reading the linked findings.
signal_count counts qualified signal sources, based on their latest qualification. Several retained findings from the same configuration can contribute only one source to that count. It is not the number of rows returned by signal history.
Source filters determine which entities match a search. signal_count and latest_signal_at still describe the entity across its visible sources. The selected evaluation can include findings with a different signal type from the source that matched the request. See Pagination and Filters for filter scope and sorting.
Dates and historical assessments
- On a Lead or Account,
found_atmeans when the entity first entered the portal’s Directory. - On a signal,
found_atmeans when that finding was saved in Plyne. History is filtered and ordered using this timestamp. latest_signal_atis the newest visible signal finding time for the entity. It may refer to a newer rejected or unclear finding, and need not equalevaluation.found_at.observed_atis the source observation timestamp attached to a historical signal when available. Older findings may havenullhere.
Evaluations and their findings
AnEvaluation has its own occurrence ID, saved time, qualification, and AI score. Its findings contains at most five visible linked findings, each with its own saved assessment and evidence. Continue with findings_cursor on signal history using evaluation_id equal to that evaluation’s ID. Evaluation list pagination is separate from linked-finding pagination.
A Lead or Account exposes its assessment only through evaluation.qualification and evaluation.ai_score. signal_sources lists the qualified source identities shown by the portal’s signal stack; signal_count uses the same source set. Different evaluations can legitimately reference the same finding.
Evidence and qualification references
qualification.references contains links attached to that exact qualification result. On a history entry, the top-level references contains evidence links for the signal finding, including supported legacy evidence sources. These fields serve different purposes and need not contain the same links.
Each reference array returns at most 20 distinct links. An empty array means no links are available in the response; it does not mean the finding has no supporting evidence.
Data coverage
Fields are returned when supported by the public response model and available for that record. These endpoints do not automatically run enrichment or fetch missing contact information. The current public models do not expose full employment histories, education, skills, verified email/phone collections, company firmographics, arbitrary custom enrichment columns, or raw provider metadata. A signal’s type can describe the criterion that found a record without returning the entire underlying profile or dataset. For example, a Skills & About signal is evidence of a match; it does not add askills[] profile field to the Lead response.